![Sample our Built Environment journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5926/TOC-Subject-Sample-2021-Built-Environment.jpg"})
Abstract
Increasingly similar corporate disclosures may fail to provide potential investors with adequate information about a company to facilitate astute, informed, and/or rational decision making. In light of this repeatedly expressed SEC concern, the current investigation uses a sample of 1,910 annual report (10-k) filings by 234 separate equity REITs from 2000-2020 to examine the impact of year-over-year linguistic similarity on the perceived information content of corporate disclosures. Consistent with the aforementioned concerns, we document a significant increase in the degree of REIT annual report similarity over time. Despite this trend, we find that innovations in annual report disclosures remain directly related to observable changes in both the operational complexity and financial position of the firm. Furthermore, these innovations engender significant market reactions including changes in options trading activity, annual report readability, and the firm’s cost of equity capital. Lastly, these findings are both robust to the inclusion of alternative information channels and are significantly more pronounced within informationally opaque market settings. In sum, we find REIT annual reports remain a value-relevant source of information for market participants.
Keywords:
Acknowledgments
Authors’ names appear in alphabetical order and senior authorship is equally shared.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Appendix A. Variable definitions.
Appendix B. Adjusted similarity score determinants.
Appendix C. More results for options trading vs. similarity scores.
Appendix D. More results for options trading vs. market reaction to annual report modifications.
Notes
1 See, for example, the SEC’s initial proposed rule governing MD&A disclosures (https://www.sec.gov/rules/proposed/33-8098.htm), as well as public comments by former SEC Chair Mary Jo White (https://www.cfo.com/accounting-2/2014/12/tackling-disclosure-overload/).
2 Specifically, within this context earnings stability means that significant insights into a REIT’s performance and future prospects are less likely to be gleaned from routine financial metrics, and more likely to be found in the strategic discussions and innovative content of annual reports.
3 While bid-ask spreads are commonly used to measure the level of informational opacity surrounding a firm, we acknowledge that they may not always provide an ideal indication. Therefore, in our robustness tests, we take into account other potential factors that could be associated with the spread measure, such as the liquidity of the firm. Specifically, we conduct regression analysis on various well-established determinants of bid-ask spreads, and then calculate the residuals to control for these factors.
4 Additional research along this dimension includes Li (Citation2010), who employs a Naïve Bayesian Machine Learning approach to unravel the information content within corporate filings and demonstrates how these statements shape market perceptions and influence decision-making processes; Campbell et al. (Citation2014), who assess the information content of mandatory risk factor disclosures in corporate filings after an SEC mandate and conclude these disclosures are instrumental in shaping perceptions of firm risk and value; and Bushee et al. (Citation2018), who investigate the multifaceted nature of linguistic complexity in corporate disclosures, and attempt to distinguish between obfuscation and information signaling.
5 We also recognize the potential impact of SEC readability requirements. However, we argue that while these requirements might standardize certain elements of report structure and language, they do not dictate the extent of year-over-year informational repetition. We encourage future researchers to delve deeper into distinguishing the effects of regulatory compliance from genuine linguistic redundancy. In this study, our primary aim is to provide a preliminary exploration into whether and how linguistic repetitiveness may serve as a proxy for information richness, while readily acknowledging that this is only a starting point for a more comprehensive investigation into the complex dynamics between textual presentation and informational value in corporate disclosures.
6 The Compustat annual database is maintained as originally reported by the firm, but the Compustat Quarterly database is updated when a firm restates its previously reported quarterly results. See Feldman et al. (Citation2010) for further discussion of this issue.
8 Prior to 2003, our sample includes the 10-k and 10-k405 filings. The 10-k405 was discontinued after 2002.
9 More specifically, we regress raw similarity scores on the first five polynomials of length and calculate the residuals as the adjusted similarity score. In our sample, REIT annual report lengths vary from as few as 9,549 words (Agree Realty Corporation in 2001) to as many as 50,654 words (Ventas, Inc. in 2011).
10 The graphs are similar if we look at adjusted similarity scores across REIT property types. For the sake of easy readability, the adjusted similarity score graphs are not reported.
11 The adjusted similarity score numbers are in line with the range reported in Brown & Tucker (Citation2011). More specifically, they estimate score ranges from -0.090 at the 25th percentile to 0.047 at the 75th percentile.
12 As a robustness test, we also replicate our main results across a broader range of alternative measurement windows. Specifically, we’ve examined one-day, two-day, four-day, and five-day cumulative abnormal returns, with the results from each alternative observation window providing very consistent findings. More specifically, while one-day window abnormal returns exhibit the weakest impact, they remain statistically significant for our variables of interest. As such, and in the interest of brevity, only 3-day (Panel A of Table 3) and 5-day window (Panel D of Table 3) results are reported.
13 We note the market reaction to such disclosures is uniquely pronounced during the period surrounding the Global Financial Crisis of 2007-09. As this era was broadly characterized by heightened levels of economic uncertainty, it is not surprising that value relevant information disclosures exhibit increased economic significance during this period.
14 See Downs (Citation1998); Elyasiani & Jia (Citation2010); Elyasiani et al. (Citation2010); Huang & Petkevich (Citation2016); Hardin III, Nagel, Roskelley and Seagraves (2017); Cline et al. (Citation2020); Gilstrap et al. (Citation2022); and Ling et al. (Citation2021) for additional discussion and analysis of the role of motivated institutional investors in monitoring firm activities, including within REITs and REIT markets.
15 We note this methodology follows that of Brown & Tucker (Citation2011). LCR is the mean decile ranking of |chg_debtdue|, |chg_lev|, and |chg_ffo|.
16 The data for these readability measures are provided by SEC Analytics Suite.
17 In untabulated results, we employ firm fixed effects instead of property fixed effects, and find that the results are qualitatively similar.
18 In untabulated results, we also investigate whether the change of auditors would have an impact on the annual report’s similarity. We find that only 3.1% of our sample year observations were characterized by REITs involved in changing auditors, while the coefficient estimates surrounding the impact of auditor changes are consistently insignificant (though they are notionally negative). This suggests auditor changes may not be a major driving force or determining factor regarding innovations in REIT annual reports.
19 The 21 bp and 24 bp increases represent the average reaction across the various specifications shown in Table 3.
20 Our results are again very consistent if we control for the readability/tone of the annual reports.
21 As additional robustness tests, in untabulated analyses we further replicate our main results based on one-day, two-day, and four-day cumulative abnormal returns and the results remain qualitatively consistent.
22 Following Naiker et al. (Citation2013), if analysts’ forecasted earnings per share for three-, four-, and five-years ahead are not reported by I/B/E/S, we estimate the as
where
is the analysts’ long term growth rate forecast reported by I/B/E/S.
23 Both Fama & French (Citation1993) three-factor and momentum factor data is obtained from https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html
24 In untabulated results, we also examine whether annual report modifications are related to changes in the cost of debt, where the cost of debt is measured as total interest expenses divided by the sum of total long-term debt and current liabilities. Consistent with the results for the implied cost of equity, we find that reduced similarity in year-over-year annual reports is directly related to larger changes in the REIT’s cost of debt.
25 Given the inherent uncertainty regarding the timing of corporate events which drive year-over-year expositional changes in each firm’s annual report, we examine three alternative options trading observation intervals: one quarter, two quarters, and the full year immediately preceding the filing and release of the REIT’s annual report.
26 More specifically, across the four model specifications reported in Panel A of Table 7, the estimated coefficient values on our focal similarity metrics dropped by 23%, 18%, 30%, and 24%, respectively, when compared to their corresponding full sample estimates in Table 3.
27 We also replicate the findings for adjusted similarity score residuals from () in . As the observed relations exhibit qualitative consistency across both similarity measures, for the sake of clarity, we once again focus our discussion on Panel B of Table 7.
28 Specifically, we identify and define informationally transparent REITs using a median split based on each REIT’s average bid-ask spread (or bid-ask spread residuals) over the quarter immediately preceding the filing date of their annual report. Firms with relatively low (high) spreads (or spread residuals) are classified as informationally transparent (opaque).
29 In untabulated results, replacing the similarity score measure residuals with raw similarity score measures produces results which are very consistent with those reported.
30 The existing body of option literature predominantly focuses on the investigation of options trading at the weekly level (e.g., Johnson & So, Citation2012; Sheng, Citation2022). Furthermore, Blau et al. (Citation2014) demonstrate that options trading holds greater return predictive power when analyzed on a weekly basis. Consequently, the current investigation delves into option market reactions with a particular emphasis on the weekly level.
31 A univariate comparison was performed on annual report innovations before and after option introductions. The results of these comparisons are in line with the multivariate analysis, which indicates a greater number of annual report modifications after option introductions compared to the level observed before options trading was readily available.